Search

The complexity of data has been increasing not only in the dimension of the data, but also in the interrelated class relationships. Pre-defined hierarchical structure of classes helps to model such relationships in a way, which divides the whole classification problem into a set of subtasks. Feature selection, known as a data preprocessing technique, aims to selecting useful feature to improve the performance of learning algorithms. However, the issue of close class relationships, especially under the same parent node in the hierarchical structure, is adverse to the feature selection. In this paper, we introduce a novel distance metric based feature selection method with large margin nearest neighbor strategy for the hierarchical classification problems. Different from other algorithms, our proposed method is to project the data into a low-dimensional feature space for learning correct metrics, and the weights of features will be indicated in the transformation matrix. A sparsity regularization and iterative optimization ensure that superior feature subsets can be selected. Experimental evaluation on several datasets with hierarchical class structure demonstrates that the proposed approach is comparable or better than some other state-of-the-art hierarchical feature selection methods. [ABSTRACT FROM AUTHOR]

In hierarchical classification learning, the hierarchical feature selection algorithm plays an important role in overcoming the curse of dimensionality. Existing hierarchical feature selection algorithms, based on the granular computing framework, all use three basic search strategies to search for similar and dissimilar classes. These strategies compute the importance of features to the global label for feature selection. However, existing methods based on the sibling strategy can only stay at the fine-grained level for feature selection, often without considering that the fine-grained level is also continuously separated from the coarse-grained level. Thus, these methods do not take into account the features hidden below the coarse granularity, resulting in the selection of a top-heavy subset of features and the loss of many important features. Therefore, this paper proposes a Hierarchical Feature Selection Based on Neighborhood Interclass Spacing From Fine to Coarse (HFSNIS) algorithm, which aims to change the feature selection to the coarse-grained hierarchy. The framework of the HFSNIS algorithm is as follows: First, each fine-grained leaf node is coarsened to the coarsest hierarchy of granularity from fine to coarse, where the non-root ancestor node is located. Next, the search for similar and dissimilar nearest neighbors is performed at the coarsest granularity hierarchy. Finally, the features are filtered using the Neighborhood Interclass Spacing model to obtain a subset of features. Therefore, this HFSNIS algorithm based on the Coarsest Search Strategy (CSS) can reselect features that were previously ignored in the fine-grained hierarchy, resulting in a better feature subset. Finally, the proposed algorithm outperforms seven state-of-the-art feature selection algorithms on six datasets. [ABSTRACT FROM AUTHOR]

Feature selection methods face new challenges in large-scale classification tasks because massive categories are managed in a hierarchical structure. Hierarchical feature selection can take full advantage of the dependencies among hierarchically structured classes. However, most of the existing hierarchical feature selection methods are not robust for dealing with the inevitable data outliers, resulting in a serious inter-level error propagation problem in the following classification process. In this paper, we propose a robust hierarchical feature selection method with a capped ℓ 2 -norm (HFSCN), which can reduce the adverse effects of data outliers and learn relatively robust and discriminative feature subsets for the hierarchical classification process. Firstly, a large-scale global classification task is split into several small local sub-classification tasks according to the hierarchical class structure and the divide-and-conquer strategy, which makes it easy for feature selection modeling. Secondly, a capped ℓ 2 -norm based loss function is used in the feature selection process of each local sub-classification task to eliminate the data outliers, which can alleviate the negative effects outliers and improve the robustness of the learned feature weighted matrix. Finally, an inter-level relation constraint based on the similarity between the parent and child classes is added to the feature selection model, which can enhance the discriminative ability of the selected feature subset for each sub-classification task with the learned robust feature weighted matrix. Compared with seven traditional and state-of-art hierarchical feature selection methods, the superior performance of HFSCN is verified on 16 real and synthetic datasets. [ABSTRACT FROM AUTHOR]